Unlocking the power of diversity in data science

Examples of data bias are all around us. In Caroline Criado Perez’s book “Invisible Women”, she details just how central data is to the world and the increasingly important decisions in areas from education to healthcare to public policy and more that are made based on data sets that don’t represent the diversity of the population. Data is power. In our industry, hiring and recruitment technology, the impact of unrepresentative data sets is clearly pronounced. People whose lives are less likely to be included in a data set are less likely to be chosen for a job. 

Was this resume screening technology taught how to read a maternity break? Was it trained to read a military deployment overseas? The importance of demographics of technology organizations can’t be discounted here. Did anyone on the team ever have a period of erratic or freelance work to care for an elderly family member? Struggle to find work after a deployment? Experience long unemployment because they couldn’t find a wheelchair accessible job? Personal experience is the lens through which we do our jobs. The people who build our hiring technology, chat-bots, education software, social media, and all the other technology we use in our lives need to represent the diversity of its users. Not just in how they look, but in who they are and how they’ve lived their lives. Technology organizations need to prioritize true diversity, not just gender and racially diverse young people from wealthy, well educated backgrounds, the typical beneficiaries of corporate diversity and inclusion efforts.

While artificial intelligence is surging in popularity at the moment, it is not new. We’ve learned that technology is only as unbiased as the teams who make it and more importantly, the data they train it on. And all technology is built on data. Chief Data Officers and other leaders in data science are in a uniquely influential position. The people they choose to hire have perhaps the most important jobs of all, building the technology we use every day. Here are the top five things Chief Data Officers and other leaders in data science can do to improve diversity in their organizations.

1. Audit hiring tools. Companies use all kinds of tools to make hiring decisions, from chat-bots to resume screening tools to AI-video assessment tools. Yet, far too often, these tools are trained on homogenous data sets. An audit of a hiring tool even found that it had identified being named “Jared” and having played high school lacrosse as indicators of strong job performance. Whether your organization is using tools to attract more candidates or to read resumes, leverage your data science expertise to measure for disparate impact. 
2. Don’t hire a “type”. Many companies follow a formula for technology hires. Some even train specialized tools (or use personality or cognitive games and tests) to identify common personality traits amongst successful employees and to identify commonalities in potential hires. Consider, though, how this might reinforce homogenous viewpoints and backgrounds. Don’t hire a “type”, rather, expand your vision of what a technology professional can look like and come from.
3. Focus on mid-senior leadership. Executive leaders set the tone for organizational commitments and culture efforts but mid-senior leaders are the ones who make these efforts happen. Companies often have strong, diverse representation at the extremes- in entry level positions and at the most senior executive levels. However, the leaders in between have huge influence. They mentor and sponsor younger employees, they identify breakout stars, they see what’s needed to make initiatives successful, and they also serve as a visual representation of what kinds of people can succeed at your company. Prioritizing diversity amongst mid-senior leadership roles will likely improve diversity in other parts of your organization. 
4. Leverage your data expertise. Data science leaders understand the importance of measuring, tracking and analyzing business efforts better than anyone else in the organization. Yet, diversity and inclusion efforts aren’t always treated like business priorities. It can be difficult to measure the impact of DEI efforts because they’re rarely measured and tracked in a meaningful way. Data science leaders can lend their expertise here. Is the company mandating a diverse candidate slate for all leadership roles? Collect and analyze data to see if it actually moved the needle on hiring more diverse candidates. Companies often lose momentum on DEI initiatives because they try to do too much at once and have a hard time figuring out what really made an impact. This is the data science department’s area expertise and would be a deeply impactful way for data leaders to support their organizations. 
5. Call out data bias. Make sure your organization understands the potential and also the limits of data. Data and the algorithms and systems trained on it are incredible tools, but they have just as much potential for bias as anything else. Data science teams already work to measure disparate impact and ensure that data sets aren’t negatively impacting algorithms. However, not everyone knows or understands this. Use your expertise to communicate with leaders on the potential and the pitfalls of data and tools built on it.

In conclusion, data science is an incredibly important function, especially with the rapid growth in popularity of AI tools built on data. These four things, along with good diversity hiring practices can help data science leaders make positive impacts not just in their organizations, but in their communities as well. Diverse technology workforces will be better equipped to create data sets that capture the complexity of the human experience and build technology that truly serves all who use it.

Looking for more information on equitable hiring? Team Meytier is deeply passionate about this subject. Do you have a question about equitable hiring you’d like us to write about? Reach out here.

Learn how Meytier uses LLMs to streamline the hiring process.

Two things your company can do to hire more equitably.